Update app.py

app.py

@@ -55,15 +55,17 @@ def generate(
     conversation = chat_history.copy()
     conversation.append({"role": "user", "content": message})
 
+    # Tokenize the input
     input_ids = tokenizer.apply_chat_template(conversation, add_generation_prompt=True, return_tensors="pt")
     if input_ids.shape[1] > MAX_INPUT_TOKEN_LENGTH:
         input_ids = input_ids[:, -MAX_INPUT_TOKEN_LENGTH:]
         gr.Warning(f"Trimmed input from conversation as it was longer than {MAX_INPUT_TOKEN_LENGTH} tokens.")
     input_ids = input_ids.to(model.device)
 
+    # Set up the streamer
     streamer = TextIteratorStreamer(tokenizer, timeout=20.0, skip_prompt=True, skip_special_tokens=True)
     generate_kwargs = dict(
-        {"input_ids": input_ids},
+        input_ids=input_ids,
         streamer=streamer,
         max_new_tokens=max_new_tokens,
         do_sample=True,
@@ -73,14 +75,20 @@ def generate(
         num_beams=1,
         repetition_penalty=repetition_penalty,
     )
+
+    # Start the generation in a separate thread
     t = Thread(target=model.generate, kwargs=generate_kwargs)
     t.start()
 
+    # Stream the output
     outputs = []
     for text in streamer:
         outputs.append(text)
         yield "".join(outputs)
 
+    # Ensure the thread is joined after completion
+    t.join()
+
 # Generate the knowledge graph HTML file
 generate_knowledge_graph()
 
@@ -129,7 +137,8 @@ demo = gr.ChatInterface(
         ["Write a Python function to reverses a string if it's length is a multiple of 4. def reverse_string(str1): if len(str1) % 4 == 0: return ''.join(reversed(str1)) return str1 print(reverse_string('abcd')) print(reverse_string('python')) "],
         ["Rectangle $ABCD$ is the base of pyramid $PABCD$. If $AB = 10$, $BC = 5$, $\overline{PA}\perp \text{plane } ABCD$, and $PA = 8$, then what is the volume of $PABCD$?"],
         ["Difference between List comprehension and Lambda in Python lst  =  [x ** 2  for x in range (1, 11)   if  x % 2 == 1] print(lst)"],
-        ["How Many R's in the Word 'STRAWBERRY' ?"],
+        ["How many hours does it take a man to eat a Helicopter?"],
+        ["Write a 100-word article on 'Benefits of Open-Source in AI research'"],
     ],
     cache_examples=False,
     type="messages",
@@ -139,4 +148,4 @@ demo = gr.ChatInterface(
 )
 
 if __name__ == "__main__":
-    demo.queue(max_size=20).launch()
+    demo.queue(max_size=20).launch(share=True)  # Set share=True for a public link